A code reproduction of VITA.
- see Mask2Former
- see IFC
- The Datasets need to be placed up the root of detertrons as follows :catalogue_desc
- You need to add the YouTubeVIS 2019 model weight of Mask2Former about Video Instance Segmentation to the weights folder as for a pre-training model.
- Based on detectron2.
- The codes are under projects/ folder, which follows the convention of detectron2.
- You can easily import our project to the latest detectron2 by following below.
- inserting VITA_1 folder